National assessment approach to sampling error estimation

Abstract

Beginning in 1969 and continuing through 1982, RTI conducted the first 12 survey years of the National Assessment of Educational Progress (NAEP). RTI’s role included developing analytic weights and providing technical support for calculating estimates and their standard errors. Ralph Folsom, PhD, led RTI’s effort to evaluate sampling error estimation methodologies appropriate to the complex NAEP design, including a comparison of the Taylor series linearization method and jackknife variance estimation methodology. Later, his work on developing Taylor series standard errors for NAEP balanced effects became a basis for the REGRESS procedure in RTI’s SUDAAN® software. This monograph provides a historical perspective to the continuing evolution of complex survey analysis methods.